Multi-round dialogue classification method based on graph convolutional neural network

• Main Authors: DU XIUJU, GUO MING, ZHANG YULUO, ZHANG YUNJU, YANG QIANG, XING MIAOMIAO, SHI HUJUN
• Format: Patent
• Language: Chinese; English
• Published: 12.11.2021
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; HANDLING RECORD CARRIERS; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS

The invention discloses a multi-round dialogue classification method based on a graph convolutional neural network. The multi-round dialogue classification method comprises the steps of 1, performing data preprocessing on an original data set; 2, constructing a graph structure; 3, preprocessing the graph structure; 4, constructing and training a graph convolutional neural network model; and 5, after model training is completed, classifying multiple rounds of conversations on a data set by using the graph convolutional neural network model. The technical problems that irrelevant interference information in multi-round dialogue texts is much and common, too much noise is introduced into an existing model, and the final classification effect of the model is affected in the prior art are solved. 本发明公开了一种基于图卷积神经网络的多轮对话分类方法，它包括：步骤1、对原始数据集进行数据预处理；步骤2、构建图结构；步骤3、对图结构进行预处理；步骤4、构建和训练图卷积神经网络模型；步骤5、在模型训练完毕之后，使用图卷积神经网络模型在数据集上对多轮对话进行分类；解决了现有技术多轮对话文本中无关干扰信息多且普遍，现有模型引入过多噪声，影响模型最终的分类效果等技术问题。